Estimation of the Orientation of the Connector for Dual-Arm Robotic Cable Manipulation

The robotic manipulation of semi-deformable linear objects (SDLOs), such as cables featuring a connector, is currently a challenge. This work addresses the problem of inserting different kinds of non-symmetric connectors once the corresponding SDLO is grasped in proximity to its end, exploiting a dual-arm robot and an RGB in-hand camera. A vision algorithm is proposed to estimate the connector orientation, and a regrasping routine is introduced if a considerable reorientation has to be performed. The method has been experimentally validated by manipulating and inserting six different types of connectors.